scholarly journals An Automated Pipeline for Dynamic Detection of Sub-Surface Metal Loss Defects across Cold Thermography Images

Sensors ◽  
2021 ◽  
Vol 21 (14) ◽  
pp. 4811
Author(s):  
Siavash Doshvarpassand ◽  
Xiangyu Wang
Keyword(s):  
Test Piece ◽  
Surface Defects ◽  
Principal Component ◽  
Specific Area ◽  
Gaussian Mixture ◽  
Metal Loss ◽  
Test Surface ◽  
Thermal Excitation ◽  
Cold Stimulation ◽  
Surface Metal

Utilising cooling stimulation as a thermal excitation means has demonstrated profound capabilities of detecting sub-surface metal loss using thermography. Previously, a prototype mechanism was introduced which accommodates a thermal camera and cooling source and operates in a reciprocating motion scanning the test piece while cold stimulation is in operation. Immediately after that, the camera registers the thermal evolution. However, thermal reflections, non-uniform stimulation and lateral heat diffusions will remain as undesirable phenomena preventing the effective observation of sub-surface defects. This becomes more challenging when there is no prior knowledge of the non-defective area in order to effectively distinguish between defective and non-defective areas. In this work, the previously automated acquisition and processing pipeline is re-designed and optimised for two purposes: 1—Through the previous work, the mentioned pipeline was used to analyse a specific area of the test piece surface in order to reconstruct the reference area and identify defects. In order to expand the application of this device over the entire test area, regardless of its extension, the pipeline is improved in which the final surface image is reconstructed by taking into account multiple segments of the test surface. The previously introduced pre-processing method of Dynamic Reference Reconstruction (DRR) is enhanced by using a more rigorous thresholding procedure. Principal Component Analysis (PCA) is then used in order for feature (DRR images) reduction. 2—The results of PCA on multiple segment images of the test surface revealed different ranges of intensities across each segment image. This potentially could cause mistaken interpretation of the defective and non-defective areas. An automated segmentation method based on Gaussian Mixture Model (GMM) is used to assist the expert user in more effective detection of the defective areas when the non-defective areas are uniformly characterised as background. The final results of GMM have shown not only the capability of accurately detecting subsurface metal loss as low as 37.5% but also the successful detection of defects that were either unidentifiable or invisible in either the original thermal images or their PCA transformed results.

Sub-surface metal loss defect detection using cold thermography and dynamic reference reconstruction (DRR)

2021 ◽  
pp. 147592172199959
Author(s):  
Siavash Doshvarpassand ◽  
Xiangyu Wang ◽  
Xianzhong Zhao
Keyword(s):  
Surface Defects ◽  
Thermal Evolution ◽  
Principal Component ◽  
The Body ◽  
Heat Diffusion ◽  
Metal Loss ◽  
Thermal Excitation ◽  
Thermal Camera ◽  
Cold Stimulation ◽  
Surface Metal

Corrosion is considered a destructive phenomenon that affects almost all metals. Active infrared thermography is an online (no result delay) and non-intrusive (no process disruption) method of non-destructive testing (NDT), which has shown profound capabilities of detecting sub-surface metal loss. However, thermal reflections, non-uniform stimulation and lateral heat diffusion will remain as the most undesirable phenomena preventing the effective observation of sub-surface defects. This becomes more challenging when there is no a priori knowledge of the anomalies to effectively distinguish between defective and non-defective areas. In this work, cooling stimulation is considered as the thermal excitation mean as 1- a very few reports in this regard have been mentioned in the body of literature and 2- a dynamic setup was achieved that is found to be effective to minimise the possibility of disrupting reflections or artefacts registered by thermal camera similar to the case of using heating stimulation. A state-of-the-art prototype mechanism was manufactured. This equipment includes a carrier carrying a thermal camera and a cooling medium reservoir operating in reciprocating motion setup. This equipment is able to scan the test piece while cold stimulation is in operation, and immediately after that the camera registers the thermal evolution. An automated contrast enhancement pipeline using a variation of adaptive histogram equalisation (AHE) combined with principal component analysis (PCA) method was developed. The enhanced image results demonstrated the capability of accurately detecting sub-surface metal loss as low as 37.5% as well as an efficiently reconstructed reference (non-defective) area.

scholarly journals Individual Tree Extraction from Terrestrial LiDAR Point Clouds Based on Transfer Learning and Gaussian Mixture Model Separation

2021 ◽  
Vol 13 (2) ◽  
pp. 223
Author(s):  
Zhenyang Hui ◽  
Shuanggen Jin ◽  
Dajun Li ◽  
Yao Yevenyo Ziggah ◽  
Bo Liu
Keyword(s):  
Transfer Learning ◽  
Gaussian Mixture Model ◽  
Mixture Model ◽  
Principal Component ◽  
Point Clouds ◽  
Gaussian Mixture ◽  
Individual Tree ◽  
Terrestrial Lidar ◽  
Initial Segmentation ◽  
Tree Extraction

Individual tree extraction is an important process for forest resource surveying and monitoring. To obtain more accurate individual tree extraction results, this paper proposed an individual tree extraction method based on transfer learning and Gaussian mixture model separation. In this study, transfer learning is first adopted in classifying trunk points, which can be used as clustering centers for tree initial segmentation. Subsequently, principal component analysis (PCA) transformation and kernel density estimation are proposed to determine the number of mixed components in the initial segmentation. Based on the number of mixed components, the Gaussian mixture model separation is proposed to separate canopies for each individual tree. Finally, the trunk stems corresponding to each canopy are extracted based on the vertical continuity principle. Six tree plots with different forest environments were used to test the performance of the proposed method. Experimental results show that the proposed method can achieve 87.68% average correctness, which is much higher than that of other two classical methods. In terms of completeness and mean accuracy, the proposed method also outperforms the other two methods.

scholarly journals Comparison of dimensionality reduction and clustering methods for SARS-CoV-2 genome

2021 ◽  
Vol 10 (4) ◽  
pp. 2170-2180
Author(s):  
Untari N. Wisesty ◽  
Tati Rajab Mengko
Keyword(s):  
Dimensionality Reduction ◽  
Dimensional Reduction ◽  
Clustering Algorithm ◽  
Sequence Data ◽  
Clustering Algorithms ◽  
Gaussian Mixture Models ◽  
Reduction Process ◽  
Principal Component ◽  
Gaussian Mixture ◽  
Clustering Methods

This paper aims to conduct an analysis of the SARS-CoV-2 genome variation was carried out by comparing the results of genome clustering using several clustering algorithms and distribution of sequence in each cluster. The clustering algorithms used are K-means, Gaussian mixture models, agglomerative hierarchical clustering, mean-shift clustering, and DBSCAN. However, the clustering algorithm has a weakness in grouping data that has very high dimensions such as genome data, so that a dimensional reduction process is needed. In this research, dimensionality reduction was carried out using principal component analysis (PCA) and autoencoder method with three models that produce 2, 10, and 50 features. The main contributions achieved were the dimensional reduction and clustering scheme of SARS-CoV-2 sequence data and the performance analysis of each experiment on each scheme and hyper parameters for each method. Based on the results of experiments conducted, PCA and DBSCAN algorithm achieve the highest silhouette score of 0.8770 with three clusters when using two features. However, dimensionality reduction using autoencoder need more iterations to converge. On the testing process with Indonesian sequence data, more than half of them enter one cluster and the rest are distributed in the other two clusters.

Surface electromyogram power spectrum in human quadriceps muscle during incremental exercise

1993 ◽  
Vol 74 (6) ◽  
pp. 2704-2710 ◽  
Author(s):  
D. Gamet ◽  
J. Duchene ◽  
C. Garapon-Bar ◽  
F. Goubel
Keyword(s):  
Principal Component ◽  
Quadriceps Muscle ◽  
Bicycle Ergometer ◽  
Linear Increase ◽  
Total Power ◽  
Test Surface ◽  
Mean Power ◽  
Mean Power Frequency ◽  
Incremental Test ◽  
Power Frequency

Spectral electromyographic (EMG) changes in human quadriceps muscles were studied to reinvestigate discrepant results concerning mean power frequency (MPF) changes during dynamic exercise. An incremental test consisting of a quasi-linear increase in mechanical power on a bicycle ergometer (for 20–100% of maximal aerobic power) was performed by forty subjects. During this test, surface EMGs from the quadriceps muscles showed that EMG total power (PEMG) increased with a curvilinear pattern for every subject, whereas MPF kinetics varied from one subject to another. PEMG changes had the same shape, which would lead to disappointing results in terms of discrimination between subjects. The ability of normalized MPF kinetics to define significant clusters of subjects was tested using a principal component analysis. This analysis led to the projection of all experiments onto a plane and revealed a relevant grouping of MPF profiles. Differences in MPF kinetics between clusters are interpreted in terms of various possibilities of balance between physiological events leading to an increase or a decrease in MPF.

scholarly journals Hyperspectral Unmixing with Gaussian Mixture Model and Low-Rank Representation

2019 ◽  
Vol 11 (8) ◽  
pp. 911 ◽  
Author(s):  
Yong Ma ◽  
Qiwen Jin ◽  
Xiaoguang Mei ◽  
Xiaobing Dai ◽  
Fan Fan ◽  
...  
Keyword(s):  
Gaussian Mixture Model ◽  
Mixture Model ◽  
Spatial Data ◽  
Principal Component ◽  
Gaussian Mixture ◽  
Low Rank ◽  
Hyperspectral Unmixing ◽  
Low Rank Representation ◽  
Synthetic Datasets ◽  
Endmember Variability

Gaussian mixture model (GMM) has been one of the most representative models for hyperspectral unmixing while considering endmember variability. However, the GMM unmixing models only have proper smoothness and sparsity prior constraints on the abundances and thus do not take into account the possible local spatial correlation. When the pixels that lie on the boundaries of different materials or the inhomogeneous region, the abundances of the neighboring pixels do not have those prior constraints. Thus, we propose a novel GMM unmixing method based on superpixel segmentation (SS) and low-rank representation (LRR), which is called GMM-SS-LRR. we adopt the SS in the first principal component of HSI to get the homogeneous regions. Moreover, the HSI to be unmixed is partitioned into regions where the statistical property of the abundance coefficients have the underlying low-rank property. Then, to further exploit the spatial data structure, under the Bayesian framework, we use GMM to formulate the unmixing problem, and put the low-rank property into the objective function as a prior knowledge, using generalized expectation maximization to solve the objection function. Experiments on synthetic datasets and real HSIs demonstrated that the proposed GMM-SS-LRR is efficient compared with other current popular methods.

scholarly journals The effects of three-dimensional defects on one-way surface plasmon propagation for photonic topological insulators comprised of continuum media

2016 ◽  
Vol 6 (1) ◽  
Author(s):  
S. Ali Hassani Gangaraj ◽  
Andrei Nemilentsau ◽  
George W. Hanson
Keyword(s):  
Surface Plasmon ◽  
Surface Defects ◽  
Three Dimensional ◽  
Metal Loss ◽  
Surface Mode ◽  
Material Interfaces ◽  
Chern Numbers ◽  
Protected Surface ◽  
Backward Propagation ◽  
Plasmon Polaritons

Abstract We have investigated one-way surface plasmon-polaritons (SPPs) at the interface of a continuum magnetoplasma material and metal, in the presence of three-dimensional surface defects. Bulk electromagnetic modes of continuum materials have Chern numbers, analogous to those of photonic crystals. This can lead to the appearance of topologically-protected surface modes at material interfaces, propagating at frequencies inside the bandgap of the bulk materials. Previous studies considered two-dimensional structures; here we consider the effect of three-dimensional defects and show that, although backward propagation/reflection cannot occur, side scattering does take place and has significant effect on the propagation of the surface mode. Several different waveguiding geometries are considered for reducing the effects of side-scattering and we also consider the effects of metal loss.

scholarly journals Structural Health Monitoring under Nonlinear Environmental or Operational Influences

2014 ◽  
Vol 2014 ◽  
pp. 1-9 ◽  
Author(s):  
Jyrki Kullaa
Keyword(s):  
Structural Health Monitoring ◽  
Health Monitoring ◽  
Linear Models ◽  
Nonlinear Models ◽  
Novelty Detection ◽  
Minimum Mean Square Error ◽  
Simulated Data ◽  
Principal Component ◽  
Gaussian Mixture ◽  
Structural Health

Vibration-based structural health monitoring is based on detecting changes in the dynamic characteristics of the structure. It is well known that environmental or operational variations can also have an influence on the vibration properties. If these effects are not taken into account, they can result in false indications of damage. If the environmental or operational variations cause nonlinear effects, they can be compensated using a Gaussian mixture model (GMM) without the measurement of the underlying variables. The number of Gaussian components can also be estimated. For the local linear components, minimum mean square error (MMSE) estimation is applied to eliminate the environmental or operational influences. Damage is detected from the residuals after applying principal component analysis (PCA). Control charts are used for novelty detection. The proposed approach is validated using simulated data and the identified lowest natural frequencies of the Z24 Bridge under temperature variation. Nonlinear models are most effective if the data dimensionality is low. On the other hand, linear models often outperform nonlinear models for high-dimensional data.

Sign in / Sign up

Export Citation Format

Share Document